preprocess data

[bellytina]

Hi, what did you do for data preprocessing? I deal with sentence-level sentiment classification on BiLSTM. The accuracy is 86% on your dataset, but 82% on the dataset downloaded from the official website. I found that all strings were converted to lowercase and special characters were deleted. Is there anything else?